The cluster details page provides tools for monitoring nodes, managing lifecycle operations, configuring networking, and accessing administrative functions.
Cluster management operates at two levels: cluster-wide operations (resize, reconfigure, delete) that affect all nodes at once, and per-node operations (rebuild, replace, power, apply settings) that target individual nodes independently.
## Access cluster details 1. In the [Gcore Customer Portal](https://portal.gcore.com), navigate to **GPU Cloud**. 2. In the sidebar, expand **GPU Clusters** and select **Bare Metal GPU Clusters**. 3. Click a cluster name to open the details page.The cluster details page shows the cluster name, current status badge, and cluster ID at the top. Below that, a header panel displays the following fields:
| Field | Description | | --------- | ------------------------------------------------------------------------------------------------------------ | | Pkey ID | InfiniBand Partition Key ID. Displayed as "-" if InfiniBand is not configured or during cluster provisioning | | OS Distro | Operating system image installed on all nodes | | Region | Data center location | | Plan | Monthly pricing plan for the cluster |The page is organized into tabs: **Overview**, **Power**, **Networking**, **Tags**, **User Actions**, and **Delete**.
## View cluster nodes
The **Overview** tab lists all nodes (servers) in the cluster. Each node entry shows the node name, flavor, assigned IP addresses, status, timestamps, cost, and other metadata. The table supports filtering by name, date range, and status.
Cluster size can be adjusted after creation by adding or removing nodes.
1. On the **Overview** tab, click **Resize Cluster**. 2. Adjust the instance count using the **+** and **-** buttons. 3. Click **Resize**.
New nodes inherit the current cluster configuration. While resizing, most management actions are unavailable; Bare Metal H100 nodes may take 15–40 minutes to provision.
Use **Reconfigure Cluster** to change the image, SSH key, or user data for the cluster and select how to apply the update.
1. Click **Reconfigure Cluster** in the top-right corner of the cluster details page. 2. In the **Update settings** step, use the dropdowns next to **Image**, **SSH Key**, and **User data** to select new values, or leave them unchanged to keep the current settings. Click **Next**. 3. In the **Select update method** step, select how to apply the changes:
| Option | Description |
| ---------------------- | ----------------------------------------------------------------------------------------------------------- |
| **Save only** | The new configuration will be used for future scale-ups or the next rebuild. Existing nodes are unaffected. |
| **Apply settings now** | Apply changes immediately to selected nodes or the entire cluster. |
| **Schedule for later** | Apply updates automatically at a specific time. Coming soon. |
4. If **Apply settings now** is selected, select **All Instances in Cluster** or **Selected Instances** under **Apply to**. Click **Reconfigure** to confirm.
When a cluster is deleted, local NVMe storage is permanently erased; file shares and object storage remain intact.
1. Navigate to the **Delete** tab on the cluster details page. 2. Click **Delete Cluster**. 3. Confirm the deletion.
Node management operations target individual servers in the cluster independently — controlling power state, reinstalling the OS, swapping out failed nodes, and syncing updated cluster configuration to specific nodes.
### Power actionsTo control a single node:
1. Locate the node in the cluster list. 2. Click the actions menu (three dots) on the node row. 3. Select the desired action: **Power on**, **Power off**, **Soft reboot**, or **Hard reboot**.
To apply power actions to multiple nodes at once, select them using the checkboxes, click **Group actions**, and select the action. Alternatively, use the **Power** tab to soft or hard reboot all cluster nodes simultaneously.
### Rebuild node
Rebuild reinstalls the OS on a node. The dialog shows the current **Image**, **SSH Key**, and **User data**. Use the dropdown next to any field to select a new value for this rebuild, or leave it unchanged to keep the current setting. All data on local storage is deleted.
1. Click the actions menu (three dots) on the node row. 2. Select **Rebuild**. 3. In the dialog, change **Image**, **SSH Key**, or **User data** as needed, or leave them to keep current values. 4. Type `Rebuild` to confirm and click **Rebuild**.To rebuild multiple nodes at once, select them using checkboxes, click **Group actions**, and select **Rebuild**. The dialog offers two modes: **Apply same settings to all Instances** or **Per Instance configuration**.
Replace deletes the current node and provisions a new one with the same configuration (flavor, image, network settings). Use this to recover a node stuck in a failed state without changing the cluster setup.
1. Click the actions menu (three dots) on the node row. 2. Select **Replace**. 3. Confirm the operation. ### Delete nodeTo permanently remove a single node from the cluster:
1. Locate the node in the cluster list. 2. Click the actions menu (three dots) on the node row. 3. Select **Delete**. 4. Confirm the deletion.When a node's cluster configuration is outdated — because the cluster was reconfigured but that node was not yet updated — a warning appears in the node details panel.
1. Click a node name in the **Overview** tab to open its details panel. 2. Click **Apply settings**. 3. Review the diff table: each row shows the **Setting**, **Current** value, and **After Apply** value. 4. Select or deselect nodes to include. 5. Type `Apply` and click **Apply**.To apply settings to multiple outdated nodes at once, select them using checkboxes in the **Overview** tab, click **Group actions**, and select **Apply settings**.
## Networking
Network interfaces are assigned to each node at cluster creation. Public and private interfaces can be added after creation. InfiniBand interfaces are managed automatically and cannot be modified.
### View interfacesThe **Networking** tab displays network interfaces for each node. Click a node name to expand its interface details.
| Type | Description | | ---------- | ----------------------------------------------------------------------- | | Public | External IP address for internet access | | Private | Internal network for communication with other cloud resources | | InfiniBand | High-speed, low-latency inter-node network for GPU-to-GPU communication |
For flavors with InfiniBand, multiple InfiniBand interfaces are created automatically (by default, eight for H100 configurations). These appear as "GPU-cluster ib-subnet" entries in the interface list. Click an interface to expand its details, including IP address and network configuration.
### Modify interfaces
To add network interfaces on a specific node:
1. Navigate to the **Networking** tab. 2. Click a node to expand its details. 3. Click **Add Interface** or **Add Sub-Interface**. 4. Configure the interface type (public or private) and IP allocation settings as supported by the service.Administrative tools for organizing clusters with tags, tracking operations through the audit log, and accessing nodes for troubleshooting via the browser console.
### TagsTags are applied at the cluster level and inherited by all nodes.
1. Navigate to the **Tags** tab on the cluster details page. 2. Enable the **Add custom tags** checkbox. 3. Enter the key and value for each tag. 4. Click **Save changes**.
Tags can be modified even while the cluster is in the Resizing state.
### User actionsThe **User Actions** tab displays a log of all operations performed on the cluster — creation, deletion, resize, power, and network actions. Use the date and action type filters to narrow results.
### Console
For troubleshooting or when SSH access is unavailable, use the browser-based console.
1. In the **Overview** tab, locate the target node. 2. Click **Open Console** in the node row. 3. The console opens in a new browser tab using noVNC. 4. Log in using the operating system credentials.
The Customer Portal is suitable for managing individual clusters. For automated workflows — CI/CD pipelines, infrastructure-as-code, or batch provisioning — the [API](#api) and [Terraform](#terraform) tabs cover the same operations programmatically.
Use the REST API to inspect clusters, resize them, manage node lifecycle operations, and automate cluster administration workflows.
Open a terminal and set these environment variables before running the examples:
```bash theme={null} export GCORE_API_KEY="{YOUR_API_KEY}" export GCORE_CLOUD_PROJECT_ID="{YOUR_PROJECT_ID}" export GCORE_CLOUD_REGION_ID="{YOUR_REGION_ID}" ```GET /cloud/v1/tasks/\{task\_id} every five seconds until `state` is `FINISHED`. Bare Metal H100 nodes may take 15–40 minutes for provisioning operations.
List all clusters in a project and region. Query string filters — by name, flavor, node count, tags, and timestamps — are documented in the [API reference](/api-reference/cloud/gpu-baremetal).
Retrieve the full configuration and current status of a single cluster.
Change the node count by specifying the new total number of nodes. Scaling down removes a random node; to target a specific node, the [delete node](#delete-a-specific-node) operation below handles per-node removal.
The `action` endpoint applies a power operation to all nodes in the cluster simultaneously.
| Action | Description | | ------------- | ------------------------------------ | | `start` | Power on all nodes | | `stop` | Power off all nodes | | `soft_reboot` | Send ACPI shutdown signal and reboot | | `hard_reboot` | Force-reset all nodes immediately |Reinstalls the OS on all nodes in the cluster using the cluster's current image, SSH key, and user data. All local storage data is erased.
Reinstalls the OS on a single node without affecting other nodes in the cluster. All local storage on the node is erased. The rebuild uses the image, SSH key, and user data currently configured in the cluster's server settings — to use different values, first update the server settings via PATCH /cloud/v3/gpu/baremetal/\{project\_id}/\{region\_id}/clusters/\{cluster\_id}/servers\_settings, then trigger the rebuild.
Delete a node and provision a new one with the same configuration. Use this to recover a node stuck in a failed state without changing the cluster setup.
Syncs a specific node that has outdated cluster configuration. The node must have `has_pending_changes=true`. To apply settings to all nodes at once, use `POST /cloud/v3/gpu/baremetal/{project_id}/{region_id}/clusters/{cluster_id}/apply_settings` without specifying a server ID.
Remove a single node by ID. If this is the last node in the cluster, the entire cluster is deleted — no confirmation prompt is shown through the API.
Delete the cluster and all its nodes. Local NVMe storage is permanently erased. File shares and object storage remain intact. Pass `all_floating_ips=true` to release any floating IPs, and `all_reserved_fixed_ips=true` to release reserved fixed IPs.
Manage the lifecycle of an existing Bare Metal GPU cluster using the [`gcore_cloud_gpu_baremetal_cluster`](https://registry.terraform.io/providers/G-Core/gcore/latest/docs/resources/cloud_gpu_baremetal_cluster) resource from the [Terraform provider](/developer-tools/terraform/overview) v2 — rename it, scale the node count, update tags, and delete it when no longer needed. To bring an existing cluster under Terraform management without recreating it, use `terraform import`.
To manage an existing cluster with Terraform without recreating it, declare the resource block with values that match the live cluster, then run `terraform import`.
```hcl theme={null} resource "gcore_cloud_gpu_baremetal_cluster" "my_cluster" { project_id = var.project_id region_id = var.region_id name = "my-cluster" # must match the current cluster name flavor = "bm3-ai-1xlarge-a100-80-8" image_id = "d9400449-0ee8-4c60-adb9-fc545a80325c" servers_count = 2 servers_settings = { interfaces = [{ type = "external" }] credentials = { ssh_key_name = "my-ssh-key" } } } ``` ```bash theme={null} terraform import gcore_cloud_gpu_baremetal_cluster.my_cluster 'Change the `name` field and run `terraform apply`. The rename completes in seconds without node restarts.
```hcl theme={null} resource "gcore_cloud_gpu_baremetal_cluster" "my_cluster" { project_id = var.project_id region_id = var.region_id name = "my-cluster-renamed" # updated name flavor = "bm3-ai-1xlarge-a100-80-8" image_id = "d9400449-0ee8-4c60-adb9-fc545a80325c" servers_count = 2 servers_settings = { interfaces = [{ type = "external" }] credentials = { ssh_key_name = "my-ssh-key" } } } ``` ## Scale the clusterChange `servers_count` and run `terraform apply`. Scaling up adds nodes; scaling down removes a random node. Clusters must always contain at least one node.
Tags are key-value pairs attached to the cluster. Update the `tags` map and run `terraform apply` — changes take effect immediately.
```hcl theme={null} resource "gcore_cloud_gpu_baremetal_cluster" "my_cluster" { project_id = var.project_id region_id = var.region_id name = "my-cluster" flavor = "bm3-ai-1xlarge-a100-80-8" image_id = "d9400449-0ee8-4c60-adb9-fc545a80325c" servers_count = 2 servers_settings = { interfaces = [{ type = "external" }] credentials = { ssh_key_name = "my-ssh-key" } } tags = { env = "production" team = "ml" } } ``` ## Delete the clusterRemove the resource block from the configuration and run `terraform apply`, or run `terraform destroy` to delete all resources at once.